This sentiment analysis has been done through the use of the “Bing” lexicon.
This lexicon was first published in:
Minqing Hu and Bing Liu, “Mining and summarizing customer reviews.”, Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (KDD-2004), Seattle, Washington, USA, Aug 22-25, 2004.
For more info on Bing: https://rdrr.io/cran/tidytext/man/sentiments.html
Looking at the sentiment bar charts one can see both the most used positive and negative words used in tweets and their contribution to the overall sentiment.
The sentiment wordcloud was built using the “BING” lexicon to categorise the overall sentiment into positive and negative words.
For more info on Bing: https://rdrr.io/cran/tidytext/man/sentiments.html
This sentiment wordcloud is a visual representation of the sentiment found in tweets regarding Covid-19. The red words represent negative sentiment and blue represent positive sentiment. The larger the word appears, the more frequently the word has been used in tweets.
This sentiment analysis has been done through the use of the “Bing” lexicon and combines the overall sentiment on a timeline based on the total daily words in tweets for both the Media Houses and the SA Government
Looking at these sentiment timelines there can be seen that the overall sentiment found in Covid-19 tweets as time progresses are significantly more negative than positive.
There can also be seen in the Government tweets timeline that the overall daily tweets has drastically declined since the 28th of May 2020. This is also reflected in the initial timeline.
This sentiment graph displays the proportion of daily sentiment in an an area plot in order to gain a clearer visual representation of the difference in sentiment. This sentiment analysis has been done through the use of the “Bing” lexicon and also displays the sentiment over a timeline.
Looking at the government area plot in comparison with the media house area plot, there can be seen that the media houses tweet sentiment are significant more negative than the government. For the entire area plot timeline, the media houses’ positive sentiment were never larger than the negative sentiment.
This sentiment analysis has been done through the use of the “Bing” lexicon. The overall sentiment is calculated by subtracting the amount of negative words from the amount of positive words in order to obtain the difference and identify which sentiment category appears the most.
By visualising and comparing each facet representing the different twitter accounts, there can be seen that the overall sentiment for all the various accounts are mostly negative.
The Government and Media House bar charts are based on the “NRC” lexicon that displays the amount of words placed under the various sentiment categories in descending order.
For more on NRC: https://rdrr.io/cran/lexicon/man/nrc_emotions.html
In both plots there can be seen that the sentiment category that contains the most word are the “positive” sentiment category. The is contradicting with the “Bing” sentiment timelines where the deduction is that the majority of words appeared to be negative.
This plot is based on the “AFINN” lexicon and displays the overall sentiment for all the various twitter accounts.The AFINN lexicon assigns words with a score that runs between -5 and 5, with negative scores indicating negative sentiment and positive scores indicating positive sentiment.
For more info on AFINN visit: http://www2.imm.dtu.dk/pubdb/pubs/6010-full.html
There can be seen that the overall sentiment is mostly negative which corresponds with plots using the BING lexicon. When looking at the three lexicons combined, there can be concluded that the NRC lexicon is differences the most in terms of the sentiment found in tweet regarding Covid-19.